Analysis of Automatic Annotation Suggestions for Hard Discourse-Level Tasks in Expert Domains

Many complex discourse-level tasks can aid domain experts in their work but require costly expert annotations for data creation. To speed up and ease annotations, we investigate the viability of automatically generated annotation suggestions for such tasks. As an example, we choose a task that is pa...

Full description

Saved in:
Bibliographic Details
Published inarXiv.org
Main Authors Schulz, Claudia, Meyer, Christian M, Kiesewetter, Jan, Sailer, Michael, Bauer, Elisabeth, Fischer, Martin R, Fischer, Frank, Gurevych, Iryna
Format Paper
LanguageEnglish
Published Ithaca Cornell University Library, arXiv.org 06.06.2019
Subjects
Annotations
Diagnostic systems
Domains
Segmentation
Task complexity
Texts
Viability
Online AccessGet full text

Cover

More Information
Summary:Many complex discourse-level tasks can aid domain experts in their work but require costly expert annotations for data creation. To speed up and ease annotations, we investigate the viability of automatically generated annotation suggestions for such tasks. As an example, we choose a task that is particularly hard for both humans and machines: the segmentation and classification of epistemic activities in diagnostic reasoning texts. We create and publish a new dataset covering two domains and carefully analyse the suggested annotations. We find that suggestions have positive effects on annotation speed and performance, while not introducing noteworthy biases. Envisioning suggestion models that improve with newly annotated texts, we contrast methods for continuous model adjustment and suggest the most effective setup for suggestions in future expert tasks.
ISSN:2331-8422